HNHacker News
TopNewBestAskShowJobs

colinchartier

1,097 karma · joined November 14, 2017

Co-founder @ https://webapp.io - a convenient solution for per-branch full-stack staging environments and automatic e2e tests.
submissionscomments
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
We'll snapshot the state of the database after you set it up once, and then re-load that snapshot every following time you need the database to exist.
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
12 is the max parallelization, so you could define more but some of them might queue.

All databases that run locally (mongo included) are supported, here's the doc page for that: https://layerci.com/docs/advanced-workflows/layerfiles-can-s...

We have a secrets dashboard, they are stored encrypted in our database (though many users have something like hashicorp vault with a central repository, with only the access key stored in our database). The secrets are only viewable by admins of your organization.

Most of our users build their docker images within LayerCI, then push to ECS by adding a write-only access key as a secret. Deployment is often done with something like Terraform or ArgoCD.

Thanks for mentioning the typo!

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
OS-level VFS wrapper (there are a few places you can put probes here, another approach might be eBPS)

We used to sell the features, but that actually didn't work super well - maybe we can add a "technical details" page for the nerds ;)

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
It takes significantly longer - we have advanced snapshotting logic to skip steps like "start the database" that you'd otherwise have to run every time you pushed.

We also auto-hibernate the instances, so you don't have to pay for a bunch of separate databases running in parallel.

Don't just listen to me though, try the free tier for yourself! A lot of our users combine LayerCI for developer tooling and vercel/netlify/GCP/AWS for deployment, they're quite complimentary.

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
We're halfway between CircleCI and Vercel - not for production hosting, but easily allowing for backends / databases to run (including a separate backend for every push)
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Yes, it's quite simple if you use something like docker compose.

If your frontend/backend are in separate repositories, I'd reference this doc page: https://layerci.com/docs/advanced-workflows/routing

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Most developers log in to LayerCI via GitHub or GitLab - so if you have a GitHub plan with SSO, you'll get SSO with Layer as well.

We even support self-hosted GitLab, so if you use a custom SSO provider there, developers could use that oauth flow for authentication + authorization.

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
There's a "CACHE" directive if you want traditional dep caching, but if you don't edit requirements.txt, a Layerfile will just entirely skip the pip install

You can "SNAPSHOT disabled" if you'd like certain steps to always run

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Firecracker is definitely heading in the same direction we did with their "diff snapshots", though they don't let you create a COW chain yet, so there's still quite a bit of development to be done on that front.
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Thanks, we're big fans of teleport - funny how small the developer tools world actually is.
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
We definitely support rust - you could even use "RUN REPEATABLE" to reuse the state from prior builds, the same way you'd build locally.

Happy to chat more and send you some swag if you try it out and write a blog post about your experience :) colin@layerci.com

You can't 'FROM vm/osx' or 'FROM vm/windows' yet, it's relatively difficult to navigate licensing as a startup unfortunately.

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
If you'd like to chat with me (CEO @ LayerCI) personally, I'm at colin@layerci.com
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
The idea is it starts as simple as a docker build (e.g., as easy or easier than GitHub actions, orbs, etc) but scales as your code/infra does.
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Will let you know if we know anyone that's looking for a consultant - do you have a website I could refer folks to?
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
> In particular, the 'correct way to solve this problem is to stand up an environment for your PR and structure your CI so you only re-run the parts that affect the particular change you want to make, rather than re-running the entire pipeline. It involves putting more thought into running your pipeline but often means the only tasks that get executed are copying a file and restarting your app

This is actually exactly what our platform helps with - instead of micromanaging this (and updating the configurations) you can rely on a friendly SaaS to do it for you.

I wouldn't call us "better" than Docker, we're more like "Take Docker and make it work better for CI/CD and staging servers"

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
One of the benefits of Layerfiles vs Dockerfiles is that those two examples will do the same thing - we monitor which files are read so you don't have to micromanage copying specific steps.

But I suppose "npm install" is a simplistic example, "npm run setup_our_ci_environment" would be a more likely step for a larger CI pipeline

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
The type of CI you're describing would be in the "simple" category in the OP - if your goal is just "run a single docker build and check if it fails" there are lots of free hosted tools you can set up (github actions, gitlab CI, etc)

There is in fact a burgeoning industry of devops consultants that set up CI pipelines for companies, I'd encourage you to consider it if you like the work!

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
We're actually in the process of getting SOC2 and pen tests ourselves - another benefit of a hosted offering is it can (eventually) integrate into your compliance system (e.g., vanta)

A lot of our customers are in fintech (payroll, banking, etc) so we've spent a lot of effort on our security model: https://layerci.com/security

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Thanks, Malthe - it's been great working with your team.
colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
If the state is broken, you can "rerun without snapshots" via the dropdown in the top right - and since future runs load the latest snapshot, they'd use a "clean" one.

For databases - usually users have a named S3 bucket and use a secret to authenticate, since we take memory snapshots, the top of your Layerfile can be "start the database and populate it from this specific anonymized dump" and then you can edit the file in S3 and re-run without snapshots if you'd like to reload it.

Here's the doc page for that: https://layerci.com/docs/advanced-workflows/layerfiles-can-s...

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Hey Alok,

RUN BACKGROUND basically does "./thecommand&" so it will stop after a restart, the onboarding example is ephemeral for simplicity.

If you want things to persist across restarts you'd have to add a systemd script or docker container the same way you'd run it in production.

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
BYO resources for CI sounds good in theory, but considering costs it's almost always a worse proposition:

- 10 hours / month of maintenance + 200 hours of setup @ $60/hr = $13k in the first year, not including any infrastructure costs.

If you just redirected that money to a hosted offering you'd get significantly more powerful servers without needing to allocate engineering / management resources - it also makes it significantly easier to push updates & monitor uptime on the host's end.

Even traditionally hosted tools like Atlassian's suite are moving to cloud for the same reason: https://www.atlassian.com/migration/journey-to-cloud

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
Hey Kurt, we're big fans of fly.io!

Happy to chat about tech sometime, I think you folks are using firecracker for a similar use-case? Drop me a line at colin@layerci.com

colinchartier··on Launch HN: LayerCI (YC S20) - Staging servers that act like (and replace) CI
With Kubernetes, a typical workflow might look like this:

1. Start kubernetes cluster

2. Build docker images

3. Deploy docker images (helm, kubectl, argocd, ...)

4(a) Run unit tests (kubectl exec -l k8s-app=web rake test)

4(b) Run e2e tests (kubectl run cypress)

4(c) Create ephemeral environment (EXPOSE WEBSITE localhost:8000)

Because we take memory snapshots after each step, you'd effectively get a fresh, fully-provisioned kubernetes cluster immediately after pushing instead of re-running all the steps every time (you'd skip to step 3) and then run 4{a,b,c} in parallel (they'd all "fork" the VM and get a separate copy of all of the resources)

Here are a few links that go into more detail:

- https://layerci.com/docs/tuning-performance/run-repeatable

- https://layerci.com/blog/ci-at-layerci/

colinchartier··on Launch HN: Seed (YC W21) – A Fully-Managed CI/CD Pipeline for Serverless
We're building something like what you describe (YC S20) - https://layerci.com - it's similar to OP but meant for standard containers instead of serverless.

TL;DR:

1. Install on GitHub https://github.com/apps/layerci/installations/new

2. Create files called 'Layerfile' to configure the pipeline

Docker Compose example for step 3: https://layerci.com/docs/examples/docker-compose

Then just point it at a docker swarm cluster or run the standard docker/ecs integration: https://docs.docker.com/cloud/ecs-integration/

colinchartier··on The Big List of Naughty Strings
The GitHub flavor of git essentially has two parts:

Repository (i.e., minimaxir/big-list-of-naughty-strings), and branch (i.e., master)

In this case, someone merged eliabieri/big-list-of-naughty-strings at master to minimaxir/big-list-of-naughty-strings at master via a pull request [1].

[1] https://github.com/minimaxir/big-list-of-naughty-strings/com...

colinchartier··on Show HN: Try out OSS webapps before installing them
Hey HN,

We built this as a little hackathon project on top of our "get a dozen staging servers" platform at https://layerci.com

The "show layer" gallery lets you create instantaneous staging servers to play around with software, as well as discussing it and submitting your own favorites for the community to use.

Submitting is currently pretty difficult, but we'll try to add any projects you submit, even if you don't have a screenshot or staging configuration for it.

colinchartier··on Show HN: Layer – Get a dozen staging servers per developer
Sort of, you can think of it as an entire socket-activated OS.

Our hypervisor does the logic though, we don't rely on the OS of the staging server itself for anything (think PXE)

colinchartier··on Show HN: Layer – Get a dozen staging servers per developer
> monorepo with many microservices

We scan the entire repository for Layerfiles, so you could, say, create a layerfile per service for unit tests, and a bigger Layerfile in the repo root that does the whole deployment and runs a few E2E tests.

> How do you run your layer file locally? They can be approximated by running a docker container and just copy/pasting the commands. We haven't built this yet because I know it will be a huge time sink to make it reliable (see, e.g., vagrant)

> Can you describe automated tests in a layer file? We tell you if any step fails (e.g., returns non-zero exit code) so you can just as easily run "./deploy.sh" as "./test.sh" - you can also start a webserver and run E2E or acceptance tests against it, for example.

> generate cross-service end to end code coverage We are trying to avoid anything that isn't very general - it's possible to run https://codecov.io/ within Layer if you want functionality like this.

colinchartier··on Show HN: Layer – Get a dozen staging servers per developer
Thank you! It took a lot of refining to find a pitch that resonated.

Let us know if you have a chance to try out the tool and have any other feedback. :)

← PreviousPage 3 of 5Next →